Hospitality AI Chatbot: Neural Network Intent Classifier & Flask API

An intelligent, context-driven conversational assistant designed for hotel web platforms. The system uses a feedforward neural network built with PyTorch and NLTK for natural language understanding (intent classification), integrated into a Flask REST API containerized with Docker.

---

📌 Features & Architecture

• Natural Language Processing (NLP): Tokenization, stemming, and Bag-of-Words vectorization using nltk.
• Deep Learning Core: Multi-layer PyTorch neural network trained on custom intent schemas (intents.json).
• RESTful Backend: Lightweight Flask service processing incoming chat payloads (POST requests) and serving contextual responses.
• Frontend-Ready: Pre-configured endpoints and JavaScript integration for client applications.
• Containerized Deployment: Reproducible setup using Docker and Docker Compose.

---

🛠️ Tech Stack

• Languages: Python 3.9+, JavaScript, HTML/CSS
• Machine Learning & NLP: PyTorch, NLTK, NumPy
• Backend & APIs: Flask, REST APIs (JSON payloads)
• DevOps & Environment: Docker, Docker Compose

---

📂 Project Structure

├── data/
│   └── intents.json        # Training corpus (tags, patterns, and responses)
├── models/
│   └── data.pth            # Trained PyTorch model weights
├── static/
│   ├── app.js              # Frontend asynchronous fetch logic
│   └── style.css           # Chat widget styling
├── templates/
│   └── base.html           # Web interface template
├── app.py                  # Flask API server & routing
├── chat.py                 # Standalone inference script
├── model.py                # Neural network architecture definition
├── nltk_utils.py           # Text preprocessing utilities
├── train.py                # Pipeline for model training and evaluation
├── Dockerfile              # Docker runtime container definition
├── docker-compose.yml      # Multi-container orchestration
├── requirements.txt        # Python package dependencies
└── README.md

---

🚀 Getting Started

Option 1: Run with Docker Compose (Recommended) Ensure you have Docker and Docker Compose installed:

# Clone the repository
git clone [https://github.com/kathyhernndez/YOUR_REPO_NAME.git](https://github.com/kathyhernndez/YOUR_REPO_NAME.git)
cd YOUR_REPO_NAME

# Build and launch the containerized application
docker-compose up -d
Navigate to http://localhost:5000 in your browser to access the chat interface. To stop the service:

docker-compose down
Option 2: Local Setup (Virtual Environment)
Clone the repository and prepare the environment:

git clone [https://github.com/kathyhernndez/YOUR_REPO_NAME.git](https://github.com/kathyhernndez/YOUR_REPO_NAME.git)
cd YOUR_REPO_NAME
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt
Train the intent classification model:

python train.py
This processes intents.json and exports the trained weights to data.pth.

Verify the model via CLI (Optional):

python chat.py
Start the Flask server:

python app.py
Open http://127.0.0.1:5000 in your browser.

---

📊 Dataset Schema (intents.json)

The chatbot's domain knowledge is defined via a structured JSON corpus:

tag: Target intent label (e.g., "booking", "check-in", "pricing").

patterns: Sample user inputs used for text vectorization and training.

responses: Curated outputs returned upon high-confidence intent classification.

---

📬 Authors & Contacts

• Katherine Hernández — Software Engineer & Applied AI
• Cristian D. Avella - Data Engineer